|
cannam@86
|
1 <html>
|
|
cannam@86
|
2
|
|
cannam@86
|
3 <head>
|
|
cannam@86
|
4 <title>Vorbisfile - Sample Crosslapping</title>
|
|
cannam@86
|
5 <link rel=stylesheet href="style.css" type="text/css">
|
|
cannam@86
|
6 </head>
|
|
cannam@86
|
7
|
|
cannam@86
|
8 <body bgcolor=white text=black link="#5555ff" alink="#5555ff" vlink="#5555ff">
|
|
cannam@86
|
9 <table border=0 width=100%>
|
|
cannam@86
|
10 <tr>
|
|
cannam@86
|
11 <td><p class=tiny>Vorbisfile documentation</p></td>
|
|
cannam@86
|
12 <td align=right><p class=tiny>vorbisfile version 1.3.2 - 20101101</p></td>
|
|
cannam@86
|
13 </tr>
|
|
cannam@86
|
14 </table>
|
|
cannam@86
|
15
|
|
cannam@86
|
16 <h1>What is Crosslapping?</h1>
|
|
cannam@86
|
17
|
|
cannam@86
|
18 <p>Crosslapping blends two samples together using a window function,
|
|
cannam@86
|
19 such that any sudden discontinuities between the samples that may
|
|
cannam@86
|
20 cause clicks or thumps are eliminated or blended away. The technique
|
|
cannam@86
|
21 is nearly identical to how Vorbis internally splices together frames
|
|
cannam@86
|
22 of audio data during normal decode. API functions are provided to <a
|
|
cannam@86
|
23 href="ov_crosslap.html">crosslap transitions between seperate
|
|
cannam@86
|
24 streams</a>, or to crosslap when <a href="seeking.html">seeking within
|
|
cannam@86
|
25 a single stream</a>.
|
|
cannam@86
|
26
|
|
cannam@86
|
27 <h1>Why Crosslap?</h1>
|
|
cannam@86
|
28 <h2>The source of boundary clicks</h2>
|
|
cannam@86
|
29
|
|
cannam@86
|
30 <p>Vorbis is a lossy compression format such that any compressed
|
|
cannam@86
|
31 signal is at best a close approximation of the original. The
|
|
cannam@86
|
32 approximation may be very good (ie, indistingushable to the human
|
|
cannam@86
|
33 ear), but it is an approximation nonetheless. Even if a sample or set
|
|
cannam@86
|
34 of samples is contructed carefully such that transitions from one to
|
|
cannam@86
|
35 another match perfectly in the original, the compression process
|
|
cannam@86
|
36 introduces minute amplitude and phase errors. It's an unavoidable
|
|
cannam@86
|
37 result of such high compression rates.
|
|
cannam@86
|
38
|
|
cannam@86
|
39 <p>If an application transitions instantly from one sample to another,
|
|
cannam@86
|
40 any tiny discrepancy introduced in the lossy compression process
|
|
cannam@86
|
41 becomes audible as a stairstep discontinuity. Even if the discrepancy
|
|
cannam@86
|
42 in a normal lapped frame is only .1dB (usually far below the
|
|
cannam@86
|
43 threshhold of perception), that's a sudden cliff of 380 steps in a 16
|
|
cannam@86
|
44 bit sample (when there's a boundary with no lapping).
|
|
cannam@86
|
45
|
|
cannam@86
|
46 <h2>I thought Vorbis was gapless</h2>
|
|
cannam@86
|
47
|
|
cannam@86
|
48 <p>It is. Vorbis introduces no extra samples at the beginning or end
|
|
cannam@86
|
49 of a stream, nor does it remove any samples. Gapless encoding
|
|
cannam@86
|
50 eliminates 99% of the click, pop or outright blown speaker that would
|
|
cannam@86
|
51 occur if boundaries had gaps or made no effort to align
|
|
cannam@86
|
52 transitions. However, gapless encoding is not enough to entirely
|
|
cannam@86
|
53 eliminate stairstep discontinuities all the time for exactly the
|
|
cannam@86
|
54 reasons described above.
|
|
cannam@86
|
55
|
|
cannam@86
|
56 <p>Frame lapping, like Vorbis performs internally during continuous
|
|
cannam@86
|
57 playback, is necessary to eliminate that last epsilon of trouble.
|
|
cannam@86
|
58
|
|
cannam@86
|
59 <h1>Easiest Crosslap</h1>
|
|
cannam@86
|
60
|
|
cannam@86
|
61 The easiest way to perform crosslapping in Vorbis is to use the
|
|
cannam@86
|
62 lapping functions with no other extra effort. These functions behave
|
|
cannam@86
|
63 identically to when lapping isn't used except to provide
|
|
cannam@86
|
64 at-least-very-good lapping results. Crosslapping will not introduce
|
|
cannam@86
|
65 any samples into or remove any samples from the decoded audio; the
|
|
cannam@86
|
66 only difference is that the transition is lapped. Lapping occurs from
|
|
cannam@86
|
67 the current PCM position (either in the old stream, or at the position
|
|
cannam@86
|
68 prior to calling a lapping seek) forward into the next
|
|
cannam@86
|
69 half-short-block of audio data to be read from the new stream or
|
|
cannam@86
|
70 position.
|
|
cannam@86
|
71
|
|
cannam@86
|
72 <p>Ideally, vorbisfile internally reads an extra frame of audio from
|
|
cannam@86
|
73 the old stream/position to perform lapping into the new
|
|
cannam@86
|
74 stream/position. However, automagic crosslapping works properly even
|
|
cannam@86
|
75 if the old stream/position is at EOF. In this case, the synthetic
|
|
cannam@86
|
76 post-extrapolation generated by the encoder to pad out the last block
|
|
cannam@86
|
77 with appropriate data (and avoid encoding a stairstep, which is
|
|
cannam@86
|
78 inefficient) is used for crosslapping purposes. Although this is
|
|
cannam@86
|
79 synthetic data, the result is still usually completely unnoticable
|
|
cannam@86
|
80 even in careful listening (and always preferable to a click or pop).
|
|
cannam@86
|
81
|
|
cannam@86
|
82 <p>Vorbisfile will lap between streams of differing numbers of
|
|
cannam@86
|
83 channels. Any extra channels from the old stream are ignored; playback
|
|
cannam@86
|
84 of these channels simply ends. Extra channels in the new stream are
|
|
cannam@86
|
85 lapped from silence. Vorbisfile will also lap between streams links
|
|
cannam@86
|
86 of differing sample rates. In this case, the sample rates are ignored
|
|
cannam@86
|
87 (no implicit resampling is done to match playback). It is up to the
|
|
cannam@86
|
88 application developer to decide if this behavior makes any sense in a
|
|
cannam@86
|
89 given context; in practical use, these default behaviors perform
|
|
cannam@86
|
90 sensibly.
|
|
cannam@86
|
91
|
|
cannam@86
|
92 <h1>Best Crosslap</h1>
|
|
cannam@86
|
93
|
|
cannam@86
|
94 <p>To acheive the best possible crosslapping results, avoid the case
|
|
cannam@86
|
95 where synthetic extrapolation data is used for crosslapping. That is,
|
|
cannam@86
|
96 design loops and samples such that a little bit of data is left over
|
|
cannam@86
|
97 in sample A when seeking to sample B. Normally, the end of sample A
|
|
cannam@86
|
98 and the beginning of B would overlap exactly; this allows
|
|
cannam@86
|
99 crosslapping to perform exactly as it would within vorbis when
|
|
cannam@86
|
100 stitching audio frames together into continuous decoded audio.
|
|
cannam@86
|
101
|
|
cannam@86
|
102 <p>The optimal amount of overlap is half a short-block, and this
|
|
cannam@86
|
103 varies by compression mode. Each encoder will vary in exact block
|
|
cannam@86
|
104 size selection; for vorbis 1.0, for -q0 through -q10 and 44kHz or
|
|
cannam@86
|
105 greater, a half-short block is 64 samples.
|
|
cannam@86
|
106
|
|
cannam@86
|
107 <br><br>
|
|
cannam@86
|
108 <hr noshade>
|
|
cannam@86
|
109 <table border=0 width=100%>
|
|
cannam@86
|
110 <tr valign=top>
|
|
cannam@86
|
111 <td><p class=tiny>copyright © 2000-2010 Xiph.Org</p></td>
|
|
cannam@86
|
112 <td align=right><p class=tiny><a href="http://www.xiph.org/ogg/vorbis/">Ogg Vorbis</a></p></td>
|
|
cannam@86
|
113 </tr><tr>
|
|
cannam@86
|
114 <td><p class=tiny>Vorbisfile documentation</p></td>
|
|
cannam@86
|
115 <td align=right><p class=tiny>vorbisfile version 1.3.2 - 20101101</p></td>
|
|
cannam@86
|
116 </tr>
|
|
cannam@86
|
117 </table>
|
|
cannam@86
|
118
|
|
cannam@86
|
119 </body>
|
|
cannam@86
|
120
|
|
cannam@86
|
121 </html>
|
